Upload
trinhtram
View
215
Download
1
Embed Size (px)
Citation preview
Project Report Water Demand Analysis and Updating Decision Support System (DSS) Models for Wyong Shire and Gosford City Councils
August 2012
The SKM logo trade mark is a registered trade mark of Sinclair Knight Merz Pty Ltd.
Project Report Water Demand Analysis and Updating Decision Support System (DSS) Models for Wyong Shire and Gosford City Councils
August 2012
Sinclair Knight Merz ABN 37 001 024 095 100 Christie Street St Leonards NSW 2065 Australia Postal Address PO Box 164 St Leonards NSW 2065 Australia Tel: +61 2 9928 2100 Fax: +61 2 9928 2500 Web: www.skmconsulting.com
LIMITATION: This report has been prepared on behalf of and for the exclusive use of Sinclair Knight Merz Pty Ltd’s Client, and is subject to and issued in connection with the provisions of the agreement between Sinclair Knight Merz and its Client. Sinclair Knight Merz accepts no liability or responsibility whatsoever for or in respect of any use of or reliance upon this report by any third party.
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE i
SINCLAIR KNIGHT MERZ
Contents 1. Introduction 1
1.1. Background 1 1.2. Challenges in Preparing Forecasts of Water Sales 1 1.3. Glossary 2
2. Demand Forecasting Approach 3
2.1. The Water and Wastewater Trend Tracking Model 3 2.2. Customer Consumption Trend Tracking 4 2.3. The DSM DSS Model 5
3. Supporting Analysis 6
3.1. Time Series Analysis of Bulk Per Capita Water Demand 6 3.2. Time Series Analysis of Bulk Per Capita Wastewater Flows 7 3.3. Comparison of Bulk Water Demand and Wastewater Flows 9 3.4. Time Series Analysis of Customer Consumption Data 11
4. Departure from Previous Forecasts 12
4.1. Impact of Population Forecast Changes and Climate 12 4.2. Impact of Demand Management and Water Restrictions 14
5. Forecasts of Bulk Demand and Consumption 19
5.1. Demand Scenarios 19 5.2. Water Consumption Forecasts 19 5.3. Forecasts for Cool, Wet Periods 21 5.4. Bulk Demand Forecasts 23 5.5. Stochastic Demand Generation 23
6. References 26
Appendix A Demand Analysis – Detailed Outline of Results 27
Appendix B Demand Management Water Savings Estimates 28
Appendix C Stochastic Generation of Demands 30
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE ii
SINCLAIR KNIGHT MERZ
Document history and status Revision Date issued Reviewed by Approved by Date approved Revision type
0 6 June 2012 Russell Beatty 6 June 2012 Preliminary Draft for client comment
1 6 July 2012 Michael Goldman
John Wall 6 July 2012 Internal Draft for Review
2 9 July 2012 Russell Beatty John Wall 9 July 2012 Draft Report
3 Russell Beatty John Wall Final Report
Distribution of copies Revision Copy no Quantity Issued to
0 1 1 Satpal Singh – Wyong Shire Council
2 1 1 Satpal Singh – Wyong Shire Council
3 1 1 Satpal Singh – Wyong Shire Council
Printed: 5 September 2012
Last saved: 5 September 2012 03:51 PM
File name: I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx
Author: Russell Beatty
Project manager: Russell Beatty
Name of organisation: Wyong Shire Council
Name of project: Water Demand Analysis and Updating Decision Support System (DSS) Models for Wyong Shire and Gosford City Councils
Name of document: Preliminary Forecasts Discussion Paper
Document version: 2
Project number: EN03189
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 1
SINCLAIR KNIGHT MERZ
1. Introduction 1.1. Background
As part of their submission to the Independent Pricing and Regulatory Tribunal of NSW (IPART) for the 2013/14-2016/17 regulatory period, both Gosford and Wyong Council water businesses are to prepare demand forecasts. The forecasts of water sales for the previous regulatory period 2009/10 to 2012/13 were well above the sales that occurred. This has led to a shortfall in consumption-based water revenue. Both Councils are eager to firstly understand why water sales in the current 2009/10 to 2012/13 regulatory period have been lower than forecast and to use that understanding to improve the accuracy of forecasts for the next regulatory period.
The last 20 years has seen a significant change in the management of urban water supplies. The move to pay for use pricing in the mid to early 1990s resulted in significant reductions in urban water demand. Additional scrutiny of water prices was provided in the form of the economic regulation of water utilities to avoid monopoly pricing outcomes for consumers. Water businesses around Australia also embarked upon demand management programs.
Much of the Australian continent entered a severe drought over the period 2004 to 2010. The so-called “Millenium Drought” also affected the water supplies on the Central Coast.
This report outlines:
1) the analysis of changes in bulk water demands and metered water sales in recent years in an attempt to provide an increased understanding of the drivers of demand; and
2) The utilisation of the information generated in the analysis to prepare forecasts of demand for the next regulatory period.
1.2. Challenges in Preparing Forecasts of Water Sales
With a severe drought since the year 2002 triggering high levels of restrictions on water use, and a comprehensive demand management effort, it is difficult to separate trends in demand from the effects of water restrictions and the impacts of efforts by both Councils and the Central Coast Community to reduce demand.
Forecasts of water sales will be dependent on understanding:
1) What is the current level of water consumption in the residential and non-residential sectors?
2) What are the impacts of the demand management programs implemented during the recent drought?
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 2
SINCLAIR KNIGHT MERZ
3) How long will the water consumption take to “rebound” following the lifting of water restrictions?
1.3. Glossary
Bulk Water Demand is assumed to refer to water passing through bulk meters and treatment facilities into the reticulation system.
Water Consumption is assumed to refer to all water passing from reticulation mains into customer’s service line and billed as usage.
Non-Revenue Water (NRW) refers to the difference between water consumption and bulk water demand. Strictly speaking a system with no metering of consumption would have 100% NRW.
Internal Use refers to water used internally in buildings and would also encompass any other water consumption that is not influenced by climate. This demand is assumed to remain unchanged by seasonal effects during the year.
External Use refers to water used externally primarily for irrigation and cooling towers, but also for swimming pools, car washing and other outdoor uses.
Demand is a generic term that refers to any of the water usage terms above.
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 3
SINCLAIR KNIGHT MERZ
2. Demand Forecasting Approach To assist in the analysis of historical water demands and the preparation of forecasts, a suite of analysis and forecasting tools has been used to examine trends in water demands and wastewater flows and to prepare demand forecasts (Figure 2-1). A description of each of the models is provided below.
Figure 2-1: Demand Analysis and Forecasting Modelling Framework
2.1. The Water and Wastewater Trend Tracking Model
The daily water demand trend tracking model was originally developed by the NSW Office of Water for providing detailed information in climate-driven variations and underlying trends in water demands and wastewater flows. This model can be used to track changes in seasonal and non-seasonal water demands and dry weather wastewater flows.
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 4
SINCLAIR KNIGHT MERZ
Figure 2-3: Example Output from Customer
Consumption Tracking Model
Figure 2-2: Tracking of Bulk Water Demand and Wastewater Flows
2.2. Customer Consumption Trend Tracking
Another key aspect of understanding the impact of both demand management programs and water restrictions is the understanding of trends in customer consumption in different customer sectors. This analysis is undertaken using the Customer Consumption Trend Tracking tool, which examines the trends in quarterly water consumption. By understanding the customer sector origin of changes in water demand, we can better understand if those changes are due to temporary behaviour changes due to restrictions or more permanent behavioural or structural changes.
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 5
SINCLAIR KNIGHT MERZ
2.3. The DSM DSS Model
The DSM DSS model is an “end-use” urban water decision support model designed for use in preparing forecasts of water demand and assessing the impact of demand management options (Figure 2-4). The model prepares baseline forecasts of water demand and wastewater generation taking account of trends in:
The propagation of water efficient fixtures and appliances;
Account formation and household size;
Employment;
Discretionary water uses and the impact of income and lifestyle factors; and
Demand management programs.
The model has a number of other economic and environmental impact functions that were not used in the preparation of demand forecasts for this project.
Figure 2-4: Structure of the DSM DSS Model
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 6
SINCLAIR KNIGHT MERZ
3. Supporting Analysis 3.1. Time Series Analysis of Bulk Per Capita Water Demand
The Daily Water Tracking Model (NSW Department of Land and Water Conservation, 2002) was used to analyse the trend in bulk per capita water demand. The results of the analysis are shown in Figure 3-1 and Figure 3-2. To assist in identifying the level of reduction in internal or non-seasonal use, a 365 day moving average of the 30th percentile of demands is also plotted. The 30th percentile is a convenient level of bulk demand that effectively accounts for the daily variations in demand that occur in cool and wet winter periods (where little external water use is expected) and is not influenced by the impact of hotter and drier conditions on bulk demand. It is taken to be representative of internal or non-seasonal water use.
The change in the non-seasonal bulk water demand can act as a barometer for the overall response to water restrictions and reductions in the level of Non-Revenue Water (NRW). It essentially is an indicator of the eagerness of the community to reduce the level of their internal water use (that is, water being discharged to the sewerage system).
The reduction in non-seasonal demand coincides with the period of water restrictions in both Gosford and Wyong, and the peak reduction in bulk demand appears to be in the order of 75 Litres/person/day (25%) in Gosford and 55 Litres/person/day (21%) in Wyong. This reduction appears to have decreased with the easing of water restrictions, but interestingly has increased again in the last 12 months.
Figure 3-1: Daily Water Tracking – Gosford
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 7
SINCLAIR KNIGHT MERZ
Figure 3-2: Daily Water Tracking – Wyong
3.2. Time Series Analysis of Bulk Per Capita Wastewater Flows
The Daily Water Tracking Model was also utilised to examine the trends in per capita wastewater flows. The results of the analysis are shown in Figure 3-3and Figure 3-4. To assist in identifying the level of reduction in dry weather flow, a 365 day moving average of the 15th percentile of flows is also plotted. As with the 30th percentile used for bulk water demands, the 15th percentile is a convenient level of wastewater flow that effectively accounts for the daily variations in dry weather flow demand that occur in dry periods (where little wet weather sewerage system infiltration). It is taken to be representative of internal or non-seasonal water use.
As with the non-seasonal water demand, the change in the non-seasonal volume of wastewater discharge can act as a surrogate for the overall response to water restrictions. It essentially is a barometer of the eagerness of the community to reduce the level of their internal water use (that is, water being discharged to the sewerage system).
The initial reduction in wastewater flows coincided with the introduction of water restrictions in both Gosford and Wyong, and the peak reduction in dry weather flows appears to be in the order of 50 Litres/person/day (20%) in Gosford and 75 Litres/person/day (23%) in Wyong.
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 8
SINCLAIR KNIGHT MERZ
Interestingly, the level of dry weather wastewater flow in both Gosford and Wyong has rebounded to almost pre-restrictions levels. This is in contrast to bulk water demand levels, which seem to be trending downward again.
Figure 3-3: Daily Wastewater Tracking – Gosford
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 9
SINCLAIR KNIGHT MERZ
Figure 3-4: Daily Wastewater Tracking – Wyong
3.3. Comparison of Bulk Water Demand and Wastewater Flows
It is a useful process in the analysis phase to examine the difference between water demands and wastewater flows. In theory, in the cooler, wetter times of the year, the difference between the two should equate to the level of water supply system leakage, which is typically about 10% of the volume of bulk water demand. The results of the analysis for Gosford and Wyong are shown in Figure 3-5and Figure 3-6. Also shown are the approximate magnitudes of the NRW as the light green bands. The difference between the two is at times much lower than expected and at times the dry weather wastewater flow is higher than the non-seasonal water use. Once source of this discrepancy could be the dry weather infiltration from groundwater or ocean water. Another could be the propagation of rainwater use which is not sourced from potable water systems but still contributes to wastewater flows.
Recommendation: The source of the smaller than anticipated difference between bulk water demand and bulk wastewater flows be investigated further. This investigation should start with the wastewater flow records.
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 10
SINCLAIR KNIGHT MERZ
Figure 3-5: Comparison of Bulk Water Demands and Wastewater Flows – Gosford
Figure 3-6: Comparison of Bulk Water Demands and Wastewater Flows – Wyong
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 11
SINCLAIR KNIGHT MERZ
3.4. Time Series Analysis of Customer Consumption Data
In Gosford and Wyong, customer water consumption data is collected from most customers on a six monthly basis. This information is aggregated into annual water consumption figures that were made available for this project from the 1999/2000 to the 2010/11 fiscal years.
To undertake a meaningful time series analysis of annual data, there needs to be a sufficient length of data set upon which to calibrate a model. As climate is the most likely cause of annual demand fluctuations outside of periods of water restrictions, it is essential that a long record of consumption is available outside of periods of water restrictions (typically ten years). This is because during water restrictions the impact of these climate influences is suppressed making it difficult to calibrate statistical models. By using data from each billing period (six months), additional seasonal influences can be observed in the data and the length of data set required for meaningful analysis can be reduced (to typically five years).
Due to the short period of annual record available outside of the period of water restrictions, it was not possible to generate a statistically significant relationship between climate influence and annual consumption in any customer sector.
Recommendation: Both Councils download and analyse trends in water consumption data on a quarterly or six monthly basis in line with billing periods.
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 12
SINCLAIR KNIGHT MERZ
4. Departure from Previous Forecasts For the previous demand forecasts submitted for the current 2009/10 to 2012/13 regulatory period, both Councils utilised the DSM DSS models that had been prepared for the IWCM Strategies. A transition from restricted demands to the medium term forecast was assumed to occur during the regulatory period.
The key drivers in demand over the current 2009/10 to 2012/13 regulatory period have been:
Population growth;
Climate;
Demand management; and
Water restrictions.
The current 2009/10 to 2012/13 regulatory period forecasts all had assumptions underlying each of these drivers. The most uncertainty in outcomes in the above four drivers is the impact of the demand management programs. The impact of climate can be assessing using the results of the water tracking models, the impact of differences in population forecasts is a relatively straightforward exercise in multiplication.
4.1. Impact of Population Forecast Changes and Climate
As a first step, the historical water sales records have been adjusted for the impact of climate and differences between population forecasts and compared with the water sales forecast by the previous DSM DSS model and those submitted to IPART for the current 2009/10 to 2012/13 regulatory period. As stated above, the forecasts submitted to IPART assumed a transition between the current levels of consumption under water restrictions to the DSM DSS model forecasts during the regulatory period. The results are shown in Figure 4-1and Figure 4-2. They show that for Wyong, differences in population forecasts (the actual population was lower than forecast) and climate (the wetter than average conditions suppressed demand) plus the fact that water restrictions were not eased as early as anticipated explain the lower than anticipated water sales. For Gosford, the persistence of water sales at levels much further below forecasts suggests that there is a different dynamic in play. It appears that the water restrictions and demand management are maintaining more downward pressure on sales than in Wyong. While Gosford experienced similar climate conditions to Wyong, demand in Gosford is significantly less sensitive to climate influences (see Section 5.3).
It is important to recognise that population forecasts are prepared based on census information that is available every five years. Preparing forecasts for a future regulatory period will require using census data that is some years old. The population forecasts used for preparing forecasts for the
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 13
SINCLAIR KNIGHT MERZ
current 2009/10 to 2012/13 regulatory period were prepared in 2005, and were based on the 2001 census data. For this study, preparing forecasts for the 2013/14 to 1016/17 regulatory period will be based on the 2006 Census. So there is some uncertainty in population outcomes for a regulatory period when the forecasts are based on census data seven or eight years prior to the commencement of the period.
Recommendation: Population forecasts for the current 2009/10 to 2012/13 regulatory period should be updated as population forecasts based on the 2011 census data are available.
Figure 4-1: Correction of Historical Water Consumption – Gosford
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 14
SINCLAIR KNIGHT MERZ
Figure 4-2: Correction of Historical Water Consumption – Wyong
Conclusion: The major difference in water sales forecasts submitted for the current 2009/10 to 2012/13 regulatory period and actual sales outcomes has been:
In Gosford, the continuation of reasonably high levels of water restrictions into the regulatory period;
In Wyong, differences in population outcomes, the influence of cooler, wetter climate conditions and the continuation of water restrictions.
Water sales in Gosford are less sensitive to climate influences that in Wyong.
4.2. Impact of Demand Management and Water Restrictions
To examine the potential impact of both Councils Demand Management efforts and the impact of water restrictions, the DSM DSS models were set up to as closely as possible replicate the impacts of the demand management programs. These programs and their forecast water savings are shown in Table 4-1. The table does not show the aggregated total water savings, because the aggregated savings need to take account of the interactions (reduced impact) of measures that target the same end uses
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 15
SINCLAIR KNIGHT MERZ
Table 4-1: Assumed Impact of Demand Management Programs
Measure Name Modelled Water Savings – Gosford (2012 - ML/a)
Modelled Water Savings – Wyong (2012 - ML/a)
National Water Efficiency Labelling Scheme (WELS)
82 83
Current Water Pricing Path 693 614
BASIX 273 401
Pre-BASIX Residential Development Controls
24 31
Residential Refit Program 72 64
Water Usage Audits of Non-Residential Properties (includes Water Plans)
251 601
All System Water Loss Programs
924 165
Storm Water Harvesting for Cricket Pitches
- 1
Public Education 1,367 1,462
Rainwater Tanks in Schools 11 1
Recycled and Groundwater User for Tankers
37 150
Rural Fire Services 5 5
Rainwater Tank Rebate – Residential
179 353
Washing Machine Rebate 38 33
Rainwater Tank for General Community Use
- 3
Groundwater 65 32
Stormwater Harvesting Projects
- 450
Rainwater Tanks for 95% New Non-Residential Development
25 -
Water Recycling Programs 187 2,408
Minor Programs 7 1
Rainwater Tanks for Council Fields and Facilities
35 4
Total Aggregate of Savings 4,275 6,862
Total Savings with Interactions 4,005 5,186
In addition, the water restrictions were assumed to impact on internal and external water uses as outlined in Table 4-2.
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 16
SINCLAIR KNIGHT MERZ
Two scenarios were tested for each Council:
1) The impact of the demand management program; and
2) The impact of the demand management program plus restrictions.
The impacts for Gosford are shown in Figure 4-3 and for Wyong in Figure 4-4. The results show that the modelling undertaken tends at times over-estimates and at other times under-estimates the reductions in water sales, with similar results for both Gosford and Wyong. One noticeable difference in the observed consumption is the increase in consumption following the easing of water restrictions in 2010. While residential consumption has increased slightly in Gosford there has been a much bigger increase in Wyong. The reason for this difference is unclear.
The exercise does highlight the difficulty in modelling consumption in two different supply areas with a consistent set of assumptions at the same time as typing to differentiate between the impact of more permanent demand management impacts and the temporary impact of water restrictions. In addition, the modelling assumes that the transition that occurs when customers move from one level of water restrictions to another is instantaneous, whereas when restrictions are easing, there will be a transition period that might make it difficult to simulate the impacts in any one year.
Table 4-2: Assumed Reductions in Water Use with Water Restrictions
Level of Restrictions
Assumed to Apply in
Year Ending June
Residential Reduction in Internal Use
Residential Reduction in External Use
Proportion of Residential Accounts
Fully Complying
Non Residential - Reduction in Internal Use
Non Residential
Reduction in External Use
Proportion of Non
Residential Accounts
Fully Complying
2.2 2005 5% 60% 50% 0% 15% 50%
3 2006 5% 60% 60% 0% 15% 60%
4 2007 5% 60% 80% 0% 20% 80%
31 2008 5% 60% 70% 0% 20% 70%
2.82 2009-2012 5% 40% 45% 0% 15% 40%
The recent drought was both lengthy (water restrictions in place for over ten years) and severe (water in surface water storages as low as 12%). In these circumstances may customers would have gone to great lengths to both conserve water (through the on-site re-use of greywater and harvesting of rainwater) and reduce their reliance on water for garden irrigation in the future (replacement of water-intensive gardens with more drought-resistant types). In addition, some
1 The second round of Stage 3 water restrictions in 2008/09 involved slightly different measures to those in 2005/06. 2 Level 2.8 water restrictions in force over the period 2008/09 to 2011/12 were considered slightly more lenient than the Level 2.2 restrictions in 2004/05.
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 17
SINCLAIR KNIGHT MERZ
customers may have abandoned their gardens or parts of their gardens altogether or the restrictions on watering may have taught them that their gardens would survive on far less water than previously thought. As a result, there is a real possibility that the level of garden watering will remain suppressed for some time, in spite of the recent lifting of water restrictions.
Conclusion: In both Gosford and Wyong cases, the assumptions utilised in the DSM DSS model to estimate the impact of demand management and water restrictions over-estimate the water savings impact of demand management and restrictions.
Figure 4-3: Impact of Demand Management and Restrictions on Metered Water
Consumption – Gosford
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 18
SINCLAIR KNIGHT MERZ
Figure 4-4: Impact of Demand Management and Restrictions on Metered Water
Consumption – Wyong
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 19
SINCLAIR KNIGHT MERZ
5. Forecasts of Bulk Demand and Consumption 5.1. Demand Scenarios
A key issue for previous consumption forecasts in both Council areas was the assumption that water restrictions would be progressively lifted during the current 2009/10 to 2012/13 regulatory period. Predictions of the outlook for water restrictions are dependent on the future climatic conditions and as such cannot be reliably predicted years in advance.
One parameter that can be estimated with some clarity is the current levels of residential and non-residential consumption. These, in concert with estimates of population growth, provide a baseline for consideration of demands. In addition, there is some validity in the estimates of the impact of the demand management programs. This will set the trajectory of the demand forecasts following the lifting of water restrictions. The only question therefore, is to the timing and duration of the transition between the two.
Work undertaken by Sydney Water suggests that the time frame for the transition to the unrestricted demand regime will be typically short (<2 years) (Sydney Water, 2012). Unfortunately, the analysis of impacts undertaken is based on the analysis of demands in a relatively short period following water restrictions, and there is a significant amount of uncertainty over the potential for rebound in the medium term (3 to 10 years).
The main issue for the medium term is that the data sets are of adequate length to pick up the short-term response to the easing and lifting of restrictions, but may be too short to pick up any medium-term transition. For both sectors, consumption will be assumed to transition between the restricted consumption and the managed demand consumption over a period of 4 years. This period of rebound is consistent with the transitions seen in Sydney following the drought of the mid 1990’s.
Conclusion: There is uncertainty on how fast the demand bounceback will be in the Central Coast in the event that water restrictions remain lifted, what the magnitude of any bounce back will be and whether the bounce back has already occurred.
5.2. Water Consumption Forecasts
For the preparation of water sales forecasts for the 2013/14 to 2016/17 regulatory period, it has been assumed that:
the starting point for projections is the 2011/12 consumption; and
water sales in both the residential and non-residential sectors will transition from current levels to the estimated managed demand forecast in the four years from 2012/13 to 2015/16.
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 20
SINCLAIR KNIGHT MERZ
The resulting Forecasts of water sales are set out in Figure 5-1 and Figure 5-2, and Table 5-1. The noticeable drop in the managed demand forecast in 2012 is the result of the commencement of permanent water saving rules in 2012.
Figure 5-1: Water Consumption Forecasts – Gosford
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 21
SINCLAIR KNIGHT MERZ
Figure 5-2: Water Consumption Forecasts – Wyong
Table 5-1: Water Sales Forecasts (ML/annum)1
Forecast 2011/12 2012/13 2013/14 2014/15 2015/16 2016/17
Gosford – Res 9,722 10,040 10,358 10,676 10,994 10,974
Gosford – Non Res 2,134 2,062 1,991 1,919 1,847 1,857
Gosford - Total 11,856 12,102 12,349 12,595 12,841 12,830
Wyong – Res 8,904 9,052 9,199 9,347 9,494 9,473
Wyong – Non Res 2,743 2,773 2,803 2,833 2,863 2,882
Wyong Total 11,647 11,825 12,002 12,180 12,357 12,355 1Forecasts from the current model. Note that these differ from the forecasts previously provided to IPART in 2008.
5.3. Forecasts for Cool, Wet Periods
One concern of all water businesses preparing forecasts of water sales over a four year regulatory period is that deviations from “average” climate conditions will results in a shortfall in revenue. The Daily Water Tracking models provide detailed estimates of the different climate conditions that can impact on seasonal consumption. These can be applied to the seasonal consumption estimates from the DSM DSS model to generate scenarios for water sales.
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 22
SINCLAIR KNIGHT MERZ
The cumulative frequency of four yearly demand factors are shown in Figure 5-3. They show the likelihood of demands occurring over a four year period other than average. The resulting demand factors for selected percentiles are shown in Table 5-2. The results show that:
The demand in Wyong is clearly more sensitive to climate than that in Gosford because the greater range over which the demand factors vary
Over a four year period, water consumption outcomes are likely to be far less impacted by the prevailing climate conditions than it will be differences in population, demand management and water restrictions outcomes.
Figure 5-3: Cumulative Frequency of Four Year Demand Factor on Seasonal Use
Table 5-2: Four Year Demand Outcomes at Different Demand Percentiles (at Pre-Restrictions Demand Levels)
Demand Percentile Gosford Wyong 0.05 -2.0% -3.9% 0.25 -1.0% -1.9% 0.50 -0.3% -0.3% 0.75 +0.6% +1.7% 0.95 +4.6% +5.7%
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 23
SINCLAIR KNIGHT MERZ
5.4. Bulk Demand Forecasts
The forecasts of bulk water demand for the regulatory period which is assumed to be the volume of water sales plus the forecast level of NRW is provided in Table 5-3.
Table 5-3: Bulk Water Demand Forecasts (ML/annum)
Forecast 2011/12 2012/13 2013/14 2014/15 2015/16 2016/17
Gosford – Sales 11,856 12,102 12,349 12,595 12,841 12,830
Gosford – NRW 938 945 953 960 967 972
Gosford – Total 12,794 13,047 13,301 13,555 13,808 13,803
Wyong – Sales 11,647 11,825 12,002 12,180 12,357 12,355
Wyong – NRW 1,249 1,255 1,261 1,268 1,274 1,280
Wyong Total 12,896 13,080 13,263 13,447 13,631 13,636
5.5. Stochastic Demand Generation
The WATHNET suite of models are used for simulating water supply headworks and distribution systems. Within the suite, support is provided for Monte Carlo analysis including generation of multi-site hydroclimatic data and probabilistic assessment of future performance, which was used for generating probabilistic demand data for use in this project.
Annual demands from the daily water tracking models hindcast were used to generate a synthetic series of demand estimates from 1970 to 2011. These demands were used as the basis for the simulation of 1,000 replicates of potential demand sequences from 2012 to 2021. Each replicate represents a potential demand outcome given observed demands in 2011. The WATHNET model output for the generation of the replicate data sets in provided in Appendix C.
The seasonal demand component in these replicates was applied to estimates of the seasonal demand from the DSM DSS model to generate a probabilistic distribution of future water sales. The results are outlined in Table 5-4 and shown in Figure 5-4 and Figure 5-5. They show the probability of the consumption in any one year exceeding the levels shown. The results provide an indication of the types of impacts that you will have from climate influences in any one year, and also that this variability will tend to increase as consumption rebounds from water restrictions.
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 24
SINCLAIR KNIGHT MERZ
Table 5-4: Total Probabilistic Water Sales Forecasts (ML/annum)
Probability of Exceedance 2011/12 2012/13 2013/14 2014/15 2015/16 2016/17
Gosford – 0.01 10,829 11,110 11,335 11,549 11,803 11,762
Gosford – 0.05 11,148 11,403 11,694 11,902 12,182 12,226
Gosford – 0.25 11,602 11,898 12,209 12,498 12,808 12,817
Gosford – 0.50 11,891 12,235 12,565 12,897 13,230 13,283
Gosford – 0.75 12,245 12,616 12,906 13,347 13,666 13,687
Gosford – 0.95 12,724 13,151 13,471 13,883 14,349 14,241
Gosford – 0.99 12,998 13,599 13,877 14,350 14,863 14,766
Wyong – 0.01 10,636 10,787 10,911 11,089 11,144 11,174
Wyong – 0.05 10,891 11,023 11,235 11,333 11,556 11,521
Wyong – 0.25 11,289 11,482 11,667 11,819 11,995 12,032
Wyong – 0.50 11,594 11,818 11,961 12,148 12,337 12,381
Wyong – 0.75 11,911 12,133 12,284 12,524 12,680 12,700
Wyong – 0.95 12,377 12,599 12,783 13,008 13,207 13,214
Wyong – 0.99 12,693 13,012 13,115 13,247 13,658 13,494
Figure 5-4: Probabilistic Sales Forecast – Gosford
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 25
SINCLAIR KNIGHT MERZ
Figure 5-5: Probabilistic Sales Forecast - Wyong
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 26
SINCLAIR KNIGHT MERZ
6. References NSW Department of Land and Water Conservation. (2002). Daily Water Demand Trend Tracking and Climate Correction Model - Version 10 User Manual. Sydney: NSW Department of Land and Water Conservation.
NSW Department of Planning. (2009). Single Dwelling Outcomes 05-08 BASIX Building Sustainability Index. Sydney: NSW Department of Planning.
Sydney Water. (2012). Sydney Water submission to IPART’s Review of prices for Sydney Water Corporation’s water, sewerage, stormwater and other services. Sydney Water.
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 27
SINCLAIR KNIGHT MERZ
Appendix A Demand Analysis – Detailed Outline of Results
The analysis of trends in water demand were supported by the analysis of trends in bulk water and wastewater demand using a multi-variable regression approach. The statistical results of the analyses are shown in Tables A-1 to A-3. Details on the use of the model and calibration procedures can be found in (NSW Department of Land and Water Conservation, 2002)
Table A-1: Soil Moisture Index Parameters
Parameter Gosford Bulk Water
Wyong Bulk Water
Gosford Bulk Wastewater
Wyong Bulk Wastewater
Rainfall multiplier 5.99 2.78 0.48 0.20 Evaporation multiplier 0.39 0.40 7.39 5.52
Evaporation power 3.26 2.55 1.00 1.00 Base flow co-efficient 0.00 0.00 0.100 0.100
Table A-2: Regression Model Calibration Output Summary
Parameter Gosford Bulk Water
Wyong Bulk Water
Gosford Bulk Wastewater
Wyong Bulk Wastewater
R Squared: 0.49 0.50 0.50 0.50 Standard Error of Y Estimate 47.97 53.21 37.89 36.99 F Statistic 444.65 270.78 1079.78 1433.20 Degrees of Freedom 1,822 1,091 4,378 4,379 Durban Watson Statistic 1.468 1.770 0.823 0.638
Table A-3: Regression Model Variable t-test Results
Parameter Gosford Bulk Water
Wyong Bulk Water
Gosford Bulk Wastewater
Wyong Bulk Wastewater
Intercept 22.9 -5.8 37.2 62.8 Soil Moisture Index -28.4 -22.9 62.0 56.0 Maximum Temperature 6.6 5.4 Not used Not used Rainfall -4.3 8.2 -7.7 Evaporation 2.3 2.3 -4.3 Weekday Index 4.3 -3.6 -2.8 Not used
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 28
SINCLAIR KNIGHT MERZ
Appendix B Demand Management Water Savings Estimates
In the process of the development of forecasts, both Gosford and Wyong Councils provided preliminary estimates of the water savings made under demand management programs. These were either replicated in the DSM DSS model, or replaced by estimates made within the DSM DSS. The assumed water savings are shown in Tables B1 and B2. These numbers are the initial water savings estimates and have been modified to provide a closer correlation with observed demand as outlined in Section Error! Reference source not found..
Table B-1: Water Savings Estimates - Gosford
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 29
SINCLAIR KNIGHT MERZ
Table B-2: Water Savings Estimates - Wyong
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 30
SINCLAIR KNIGHT MERZ
Appendix C Stochastic Generation of Demands GENERATED DATA SUMMARY (Version 4.01) Generated data file: ../../dem/gos_wyo_dem.gen Historic data file: ../../dem/gos_wyo_dem.bin Estimation of Lag-one Annual Data Model Parameters The multi-site markov model has the form for year t: x(t) = a*x(t-1) + e(t) where x(t) is a vector of centralized transformed data obtained using the Box-Cox transformation x = (q**lambda-1)/lambda where q is the observed data, and e(t) is a disturbance vector distributed normally with zero mean and covariance sigma Std dev/correlation matrix of disturbances: ---------------------------------------------------------------------------------------------------------------------------------- 0.33936E-01 0.92915 0.50577E-01 There were NO missing data Full-data maximum likelihood estimators used **************************************************************************************************** Site name: GOSFORD (L/d) Mean of transformed data = 11.617 Disturbance std dev = 0.034 Box-Cox lambda = 0.000 Prob[negative flow] = 0.0000 Annual data prior to generation set to 105627.000 In 41 years there were 0 missing years Box-Cox lambda Skew of transformed data ---------------------------------------- 1.500 1.1233 1.450 1.1142 1.400 1.1052 1.350 1.0961 1.300 1.0871 1.250 1.0781 1.200 1.0691 1.150 1.0602 1.100 1.0512 1.050 1.0423 1.000 1.0334 0.950 1.0245 0.900 1.0157 0.850 1.0068 0.800 0.9980 0.750 0.9892 0.700 0.9804 0.650 0.9716 0.600 0.9629 0.550 0.9541 0.500 0.9454 0.450 0.9367 0.400 0.9280 0.350 0.9194 0.300 0.9108 0.250 0.9021 0.200 0.8935 0.150 0.8850 0.100 0.8764 0.050 0.8679 0.000 0.8594 Autocorrelation function Lag Autocorrelation -1.0 -.8 -.6 -.4 -.2 0.0 .2 .4 .6 .8 1.0 Estimate 95% limits on white noise |....|....|....|....|....|....|....|....|....|....| 1 -0.003 0.316 -0.316 | < * > | 2 0.032 0.320 -0.320 | < |* > | 3 0.211 0.324 -0.324 | < |***** > | 4 -0.167 0.329 -0.329 | < ****| > | 5 -0.123 0.333 -0.333 | < ***| > | 6 -0.010 0.338 -0.338 | < * > | 7 0.051 0.343 -0.343 | < |* > | 8 -0.011 0.348 -0.348 | < * > | 9 0.116 0.354 -0.354 | < |*** > | 10 -0.140 0.359 -0.359 | < ***| > | 11 0.249 0.365 -0.365 | < |****** > | 12 -0.016 0.371 -0.371 | < * > | 13 -0.252 0.378 -0.378 | < ******| > | 14 0.063 0.385 -0.385 | < |** > | 15 0.072 0.392 -0.392 | < |** > | |........................|........................| Note < and > denote approximate 95% limits on the white noise autocorrelation function
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 31
SINCLAIR KNIGHT MERZ
Partial autocorrelation function Lag Partial autocorrelation -1.0 -.8 -.6 -.4 -.2 0.0 .2 .4 .6 .8 1.0 Estimate 95% limits on white noise |....|....|....|....|....|....|....|....|....|....| 1 -0.003 0.316 -0.316 | < * > | 2 0.032 0.320 -0.320 | < |* > | 3 0.211 0.324 -0.324 | < |***** > | 4 -0.173 0.329 -0.329 | < ****| > | 5 -0.144 0.333 -0.333 | < ****| > | 6 -0.042 0.338 -0.338 | < *| > | 7 0.150 0.343 -0.343 | < |**** > | 8 0.023 0.348 -0.348 | < |* > | 9 0.076 0.354 -0.354 | < |** > | 10 -0.242 0.359 -0.359 | < ******| > | 11 0.318 0.365 -0.365 | < |********> | 12 -0.049 0.371 -0.371 | < *| > | 13 -0.193 0.378 -0.378 | < *****| > | 14 -0.140 0.385 -0.385 | < ****| > | 15 0.282 0.392 -0.392 | < |******* > | |........................|........................| Note < and > denote approximate 95% limits on the white noise partial autocorrelation function Cumulative periodogram (assumes constant error variance) ........................................... 1.000 + + +. . . . + . . + . . + . 0.764 + . . . . . . + . . + . 0.527 + . . . . + + + . . + + . . + . 0.291 . . + . . + . . + . .+ . !.........!.........!.........!.........!... 0.0 0.1 0.3 0.4 0.5 Frequency Hypothesis: errors are time-independent - test statistic = 0.0924 - 5% Exceedance value of test statistic = 0.3041 Plot of standardized residuals against time ........................................................................................... 3.504 + . . . . . . . . + . 2.027 . . . . + . . + + + . . + + + . 0.551+ + + + + + + . .--+-----------------------+---------------------------------------+----+------------------. . + + + + + . . + + + + + + . . + + + + . -0.926 + + . . + +. . + . . + . !.........!.........!.........!.........!.........!.........!.........!.........!........... 0.0 4.5 9.0 13.5 18.0 22.5 27.0 31.5 36.0 Runs test Z-statistic = 0.194
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 32
SINCLAIR KNIGHT MERZ
Plot of standardized residuals vs N(0,1) variate ............................................. 3.504 +. . . . . . . . . 2.168 + . . . . . . + . . ++ + . 0.832 +++ . . +2++2 . . +2 . .--------------------+++---------------------. . +2+2+ . -0.504 +2+ . . +++ . . + + . . + + . . . -1.840+ . !.........!.........!.........!.........!..... -2.1 -1.1 -0.1 0.8 1.8 Normal variate Hypothesis: Errors are normally-distributed - test statistic = 0.0827 - 5% Exceedance value of test statistic = 0.1371 **************************************************************************************************** Site name: WYONG (L/d) Mean of transformed data = 11.771 Disturbance std dev = 0.051 Box-Cox lambda = 0.000 Prob[negative flow] = 0.0000 Annual data prior to generation set to 120649.000 In 41 years there were 0 missing years Box-Cox lambda Skew of transformed data ---------------------------------------- 1.500 0.4482 1.450 0.4405 1.400 0.4328 1.350 0.4252 1.300 0.4175 1.250 0.4098 1.200 0.4022 1.150 0.3946 1.100 0.3869 1.050 0.3793 1.000 0.3717 0.950 0.3641 0.900 0.3566 0.850 0.3490 0.800 0.3415 0.750 0.3339 0.700 0.3264 0.650 0.3189 0.600 0.3114 0.550 0.3039 0.500 0.2964 0.450 0.2889 0.400 0.2815 0.350 0.2740 0.300 0.2666 0.250 0.2592 0.200 0.2518 0.150 0.2444 0.100 0.2370 0.050 0.2296 0.000 0.2222 Autocorrelation function Lag Autocorrelation -1.0 -.8 -.6 -.4 -.2 0.0 .2 .4 .6 .8 1.0 Estimate 95% limits on white noise |....|....|....|....|....|....|....|....|....|....| 1 -0.004 0.316 -0.316 | < * > | 2 0.054 0.320 -0.320 | < |* > | 3 0.065 0.324 -0.324 | < |** > | 4 -0.110 0.329 -0.329 | < ***| > | 5 -0.160 0.333 -0.333 | < ****| > | 6 -0.126 0.338 -0.338 | < ***| > | 7 0.028 0.343 -0.343 | < |* > | 8 -0.108 0.348 -0.348 | < ***| > | 9 -0.010 0.354 -0.354 | < * > | 10 -0.035 0.359 -0.359 | < *| > | 11 0.233 0.365 -0.365 | < |****** > | 12 -0.034 0.371 -0.371 | < *| > | 13 -0.189 0.378 -0.378 | < *****| > | 14 0.118 0.385 -0.385 | < |*** > | 15 -0.001 0.392 -0.392 | < * > | |........................|........................| Note < and > denote approximate 95% limits on the white noise autocorrelation function
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 33
SINCLAIR KNIGHT MERZ
Partial autocorrelation function Lag Partial autocorrelation -1.0 -.8 -.6 -.4 -.2 0.0 .2 .4 .6 .8 1.0 Estimate 95% limits on white noise |....|....|....|....|....|....|....|....|....|....| 1 -0.004 0.316 -0.316 | < * > | 2 0.054 0.320 -0.320 | < |* > | 3 0.066 0.324 -0.324 | < |** > | 4 -0.113 0.329 -0.329 | < ***| > | 5 -0.172 0.333 -0.333 | < ****| > | 6 -0.127 0.338 -0.338 | < ***| > | 7 0.061 0.343 -0.343 | < |** > | 8 -0.083 0.348 -0.348 | < **| > | 9 -0.041 0.354 -0.354 | < *| > | 10 -0.094 0.359 -0.359 | < **| > | 11 0.234 0.365 -0.365 | < |****** > | 12 -0.041 0.371 -0.371 | < *| > | 13 -0.270 0.378 -0.378 | < *******| > | 14 0.040 0.385 -0.385 | < |* > | 15 0.118 0.392 -0.392 | < |*** > | |........................|........................| Note < and > denote approximate 95% limits on the white noise partial autocorrelation function Cumulative periodogram (assumes constant error variance) ........................................... 1.000 + +. . + . . . . + . . + + . 0.756 + . . . . + . . + . . + . 0.513 . . + + + . . + . . + + . . . 0.269 + . . + . . . . + . .+ . !.........!.........!.........!.........!... 0.0 0.1 0.3 0.4 0.5 Frequency Hypothesis: errors are time-independent - test statistic = 0.0836 - 5% Exceedance value of test statistic = 0.3041 Plot of standardized residuals against time ........................................................................................... 2.522 + + . . . . . . + . . . 1.322 + + . . . . + + + + . . + + + + + . . + + + + + + . 0.123+----------------------------------+-------------------+----------------------------------. . + + + . . + + . . + + + + + + . . + . -1.077 + + +. . + . . . . + + . !.........!.........!.........!.........!.........!.........!.........!.........!........... 0.0 4.5 9.0 13.5 18.0 22.5 27.0 31.5 36.0 Runs test Z-statistic = -0.442
I:\ENVR\Projects\EN03189\Deliverables\Reports\EN03189 - Gosford and Wyong Water Sales Forecasts - Report Rev03.docx PAGE 34
SINCLAIR KNIGHT MERZ
Plot of standardized residuals vs N(0,1) variate ............................................. 2.522 + +. . . . . . . . + . 1.437 + + . . . . ++ . . 2++ . . +++ . 0.351 +2++ . .--------------------2++---------------------. . +2 . . + . . +2+++ . -0.734 ++ . . . . + ++ . . + . . . -1.820+ + . !.........!.........!.........!.........!..... -2.1 -1.1 -0.1 0.8 1.8 Normal variate Hypothesis: Errors are normally-distributed - test statistic = 0.0668 - 5% Exceedance value of test statistic = 0.1371 Data generation option summary: Site Name Code Group Ksite kNN Ksite2 Cl_ch Year ------------------------------------------------------------------------- 1 GOSFORD (L/d) Generate 1 y 1 1 0 42 2 WYONG (L/d) Generate 1 n 2 0 42 >>> NO parameter uncertainty Replicate Annual Data Statistics Number of replicates = 2022 Number of years = 10 Statistic Site name Mean Std dev Skew 5% 25% 50% 75% 95% -------------------------------------------------------------------------------------------------------------------------------------- Mean GOSFORD (L/d) 111036.591 1235.707 0.006 109042.082 110155.912 111041.373 111865.646 113054.968 Mean WYONG (L/d) 129635.327 2286.517 0.096 126057.865 128052.728 129629.655 131184.522 133445.277 Std dev GOSFORD (L/d) 3681.430 892.049 0.365 2312.096 3048.424 3602.185 4253.429 5245.789 Std dev WYONG (L/d) 6357.158 1535.255 0.353 3982.082 5252.890 6266.662 7367.182 8934.865 Skew GOSFORD (L/d) 0.063 0.497 0.007 -0.776 -0.250 0.061 0.393 0.878 Skew WYONG (L/d) 0.068 0.503 0.026 -0.769 -0.243 0.063 0.379 0.914 Lag-1 cor GOSFORD (L/d) -0.097 0.321 0.016 -0.620 -0.325 -0.102 0.139 0.433 Lag-1 cor WYONG (L/d) -0.054 0.318 0.061 -0.566 -0.283 -0.052 0.169 0.475